How to preprocess face before training?

asked 2018-03-31 06:31:53 -0600

nirmalkumark gravatar image

I am using haarcascade classifier for detecting faces and training them which will be used for recognition later. How to preprocess the images like smoothing, denoising etc before training. Kindly suggest me some codes in python.

Thanks in advance

edit retag flag offensive close merge delete


to avoid the XY problem, mind explaining the purpose of your program ?

(do you really want or need a "one out of x" identification task ?)

berak gravatar imageberak ( 2018-03-31 09:36:11 -0600 )edit

I just want to increase the recognizing accuracy! For that I think, better the training better will be the results. for better training i wish to do some preprocessing of the images. how to do it! I am using opencv python

nirmalkumark gravatar imagenirmalkumark ( 2018-03-31 09:52:24 -0600 )edit
  • "I just want to increase the recognizing accuracy!" -- for what purpose ,again ? different needs require differrent solutions. (e.g. if you're trying "is that me ?", it might be the wrong approach to use opencv's face recognition classes, which are doing something else)

  • "I am using opencv python" -- irrelevant

berak gravatar imageberak ( 2018-03-31 10:30:51 -0600 )edit

Bro! Actually I trained my face recognition system with 10 images each of 5 persons using Haar classifier. When i test the system by providing some 5 test images it recognises person A as person B. How to improve the accuracy! Thats my question

nirmalkumark gravatar imagenirmalkumark ( 2018-03-31 11:04:18 -0600 )edit

"When i test the system by providing some 5 test images it recognises person A as person B" - This is recognition I guess. It recognises wrongly

nirmalkumark gravatar imagenirmalkumark ( 2018-03-31 11:09:09 -0600 )edit